From: Kaxon | #283
I have thought about that before, but it's pretty random - sometimes ihatethisCPU (or someone else) makes a totally insane prediction, so it's equivalent to not getting anything. Other times it's not as bad.
In PPC, we had the same sort of issue with the weird predictors - Wigs decided to use the same percentage for every single result for every match. We ended up compromising by asking Wigs if he was ok with his predictions being ignored when giving make-up points.
You could probably pull off something similar in the Oracle - make it so that, when assigning those make-up points, only serious predictors count. You would, of course, have to ask the "non-serious" predictors and the ones who aim for top 5 picks like red sox and Ringworm if they'd be ok with that. You would probably also want to ignore predictions from people who aren't in the running - if someone starts predicting halfway through the contest and does a miserable job, it's not like they have a chance to catch up anyway.
--
thelefty for analysis crew 2008 imo -tranny
I have a third degree burn in flame-o-nomics -Sir Chris